Six Centimetres of Silver Into a Gap I Couldn't See

The pannier racks I’d been looking at were all stamped steel with zinc coating, mounting holes in the wrong places for the leather tool bag I’d stitched three days earlier. Someone in the forum mentioned custom touring frames with asymmetric geometry—head tube angle adjusted 0.8° to compensate for loaded weight distribution. The photos showed polished stainless lugs with hand-filed cutouts, the brazing so clean it looked like the tubes had grown together naturally. Apparently there’s a framebuilding co-op three blocks from the pottery studio. Torch time costs eighteen dollars per hour.

Showed up at 9:40 AM with a notebook. The instructor—Marc, maybe fifty, wearing a leather apron scorched black in streaks—was finishing a seat lug on someone else’s frame, heating it with an oxyacetylene torch while feeding silver rod into the joint. The rod disappeared into the gap between tube and lug like water into sand. No dripping, no pooling. Just gone.

“Capillary action,” he said when he noticed me watching. “You’re not filling the joint. You’re letting physics pull the braze where it needs to go.”

Started with tube mitering. A head tube meets the top tube and down tube at specific angles—73° for the head angle, which determines how the bike steers. Cut the tubes on a horizontal bandsaw with a jig that holds them at the correct angle. The cut surface needs to mate flush against the head lug’s interior socket. Gap larger than 0.08 millimetres and the braze won’t flow properly. Smaller than 0.03 and it won’t penetrate at all.

Filed the tube ends by hand until they seated cleanly into the lug, checking the fit by feel. A proper miter contacts the entire inner surface with light friction, no rocking, no gaps visible when you hold it up to the light. Got it right on the fourth attempt. The filing removed maybe two millimetres of steel over twenty minutes of work. Precision this tight makes knife sharpening look casual.

The jig is structural origami. Four adjustable clamps hold tubes to half-millimetre tolerances while you apply 650°C heat that wants to warp everything. Marc positioned the head lug assembly—top tube and down tube both mitered and inserted—then used a plumb bob and a tape measure to verify alignment. Head tube vertical within 0.3°, tube centerlines offset exactly 42 millimetres for a 56-centimetre frame. He tightened the clamps and said, “This is where you discover whether you mitered correctly. If the tubes fight the jig, your angles are wrong.”

Mine fought slightly. The down tube wanted to sit 0.7 millimetres off-center. Marc said we could shim it or re-miter. We re-mitered. Another fifteen minutes of filing until the tube dropped into place without tension.

Flux chemistry next. White paste that smells like the borax I used for heat-treating steel, but the behaviour is counterintuitive. Flux doesn’t help the braze stick. It removes surface oxides and protects hot metal from oxidation while the silver flows. When flux bubbles and turns clear like molten glass, that’s the temperature window—620° to 650°C. Add the braze rod then.

I added it too early. Flux still white and pasty, not glassy. The rod touched the tube and just sat there, not wetting, looking like a silver wire pressed against steel. Marc said, “You’re contaminating the joint. Clean it and start again.”

Cleaned with acetone. Reapplied flux. Heated until the paste bubbled, frothed, then went clear and transparent. The lug glowed faint orange—not red, Marc said red means you’ve overshot and started damaging the tube’s temper. Touched the silver rod to the joint at that exact moment. The rod melted and vanished into the gap, pulled by capillary forces I couldn’t see. Fed another three centimetres of rod. It kept disappearing. The fillet around the joint edge built up slowly, smooth and concave, following the lug’s geometry.

Stopped heating. The silver froze solid in maybe two seconds, colour shifting from molten-bright to dull grey. Marc inspected the fillet, pointed to a spot near the bottom where the braze hadn’t fully penetrated—visible as a gap in the fillet where the silver should have been continuous. “Tube was cooler there. Heat distribution matters as much as total temperature.”

Tried the lower head lug next. Heated more evenly this time, rotating the torch to keep the temperature balanced. Fed the rod around the entire perimeter in one continuous motion, watching the silver flow ahead of the rod tip, chasing the heat. Got a complete fillet with no gaps. Marc said it looked acceptable for a first attempt but pointed out three places where I’d added too much rod—the fillet sat proud of the lug edge instead of blending smoothly. “You can file those down, but better to not put it there in the first place.”

The jig held the partial frame assembly while I stared at the joints I’d just brazed. They looked permanent in a way pottery trimming doesn’t—you can always center more clay on the wheel. This tube is now part of this lug unless I reheat the joint to 620°C and melt it apart again, which is called sweating out. Repairability built into the technique, but only if you’re willing to reverse the process entirely.

Left after ninety minutes with instructions to come back and finish the seat tube and bottom bracket joints. The frame’s locked in the jig until then, clamped at angles I spent forty minutes achieving. Marc said frame geometry is cumulative error—every misaligned joint adds up, and by the time you’ve brazed the last lug, you’re either within tolerance or starting over.

The leather bag still won’t fit in the pannier. Building a custom frame to hold a custom bag to carry fly fishing equipment seems like five hobbies stacked in a trench coat pretending to be practical.